How to use 8bit character sets?

[copx]

For some reason Python (on Windows) doesn't use the system's default
character set and that's a serious problem for me.
I need to process German textfiles (containing umlauts and other > 7bit
ASCII characters) and generally work with strings which need to be processed
using the local encoding (I need to display the text using a Tk-based GUI
for example). The only solution I managed to find was converting between
unicode and latin-1 all the time (the textfiles aren't unicode, the output
of the program isn't supposed to be unicode either). Everything worked fine
until I tried to run the program on a Windows 9x machine.. It seems that
Python on Win9x doesn't really support unicode (IIRC Win9x doesn't have real
unicode support so that's not suprising).
Is it possible to tell Python to use an 8bit charset (latin-1 in my case)
for textfile and string processing by default?

copx

 

[Chris Curvey]

Check out sitecustomize.py.

http://diveintopython.org/xml_processing/unicode.html

 

[copx]

Check out sitecustomize.py.

http://diveintopython.org/xml_processing/unicode.html

Thanks but I'm looking for a way to do this on application level (i.e. I
want my app to run in an unmodified interpreter enviroment).

copx

 

[John Machin]

For some reason Python (on Windows) doesn't use the system's default
character set and that's a serious problem for me.
I need to process German textfiles (containing umlauts and other > 7bit
ASCII characters) and generally work with strings which need to be processed
using the local encoding (I need to display the text using a Tk-based GUI
for example). The only solution I managed to find was converting between
unicode and latin-1 all the time (the textfiles aren't unicode, the output
of the program isn't supposed to be unicode either). Everything worked fine
until I tried to run the program on a Windows 9x machine.. It seems that
Python on Win9x doesn't really support unicode (IIRC Win9x doesn't have real
unicode support so that's not suprising).
Is it possible to tell Python to use an 8bit charset (latin-1 in my case)
for textfile and string processing by default?

copx

1. Your description of your problem is extremely vague. If you were to
supply a minimal script that "works" [on what platform?? what version of
Python??], with a description of what you understand by "works", and
what happens differently when you run that script on a Win9x box [for
what value(s) of x?? what version of Python??], we might be able to help
you. N.B. somewhere near the top of the script you should have something
like:

import sys
print "Python version:", sys.version
print "platform:", sys.platform
print "default encoding:", sys.getdefaultencoding()
try:
print "Windows version:", sys.getwindowsversion()
except AttributeError:
print "sys.getwindowsversion not available"

2. You should read this:

http://www.catb.org/~esr/faqs/smart-questions.html

3. You should not rely on a crutch like a default encoding, especially
one obtained by a kludge like sitecustomize.py. If your app expects to
receive data in encoding x and send data in encoding y, these facts are
properties of the application and the data, NOT the box you are running
on. If you had a requirement to read MacCyrillic from a Classic Mac and
write KOI8 for consumption on a Windows PC, you should be able to do it
on a SPARC Solaris box in Timbuktu or Walla Walla, Wa., without having
to fiddle with site-wide configuration.

4. AFAIK, support for Unicode is provided by Python with no assistance
from the operating system. The multitudinous deficiencies in Win9x
should have no bearing on the problem. Have you tried to run your
program on a Win2K or WinXP box?

HTH,

John

 

[=?ISO-8859-1?Q?=22Martin_v=2E_L=F6wis=22?=]

For some reason Python (on Windows) doesn't use the system's default
character set and that's a serious problem for me.

I very much doubt this statement: Python does "use" the system's default
character set on Windows. What makes you think it doesn't?

Is it possible to tell Python to use an 8bit charset (latin-1 in my case)
for textfile and string processing by default?

That is the default.

Regards,
Martin

 

[John Roth]

I very much doubt this statement: Python does "use" the system's default
character set on Windows. What makes you think it doesn't?


That is the default.

As far as I can tell, there are actually two defaults, which tends
to confuse things. One is used whenever a unicode to 8-bit
conversion is needed on output to stdout, stderr or similar;
that's usually Latin-1 (or whatever the installation has set up.)
The other is used whenever the unicode to 8-bit conversion
doesn't have a context - that's usually Ascii-7.

John Roth

 

[=?ISO-8859-1?Q?=22Martin_v=2E_L=F6wis=22?=]

As far as I can tell, there are actually two defaults, which tends
to confuse things.

Notice that there are two defaults already in the operating system:
Windows has the notion of the "ANSI code page" and the "OEM code
page", which are used in different contexts.

One is used whenever a unicode to 8-bit
conversion is needed on output to stdout, stderr or similar;
that's usually Latin-1 (or whatever the installation has set up.)

You mean, in Python? No, this is not how it works. On output
of 8-bit strings to stdout, no conversion is ever performed:
the byte strings are written to stdout as-is.

The other is used whenever the unicode to 8-bit conversion
doesn't have a context - that's usually Ascii-7.

Again, you seem to be talking about Unicode conversions -
it's not clear that the OP is actually interested in
Unicode conversion in the first place.

Regards,
Martin

 

[John Roth]

Notice that there are two defaults already in the operating system:
Windows has the notion of the "ANSI code page" and the "OEM code
page", which are used in different contexts.


You mean, in Python? No, this is not how it works. On output
of 8-bit strings to stdout, no conversion is ever performed:
the byte strings are written to stdout as-is.

That's true, but I was talking about outputing unicode strings,
not 8-bit strings. As you say below, the OP may not have
been talking about that.

Again, you seem to be talking about Unicode conversions -
it's not clear that the OP is actually interested in
Unicode conversion in the first place.

Regards,
Martin

John Roth